Journal of Applied Physiology

Bicarbonate attenuates arterial desaturation during maximal exercise in humans

Henning B. Nielsen, Per P. Bredmose, Morten Strømstad, Stefanos Volianitis, Bjørn Quistorff, Niels H. Secher


The contribution of pH to exercise-induced arterial O2 desaturation was evaluated by intravenous infusion of sodium bicarbonate (Bic, 1 M; 200–350 ml) or an equal volume of saline (Sal; 1 M) at a constant infusion rate during a “2,000-m” maximal ergometer row in five male oarsmen. Blood-gas variables were corrected to the increase in blood temperature from 36.5 ± 0.3 to 38.9 ± 0.1°C (P < 0.05; means ± SE), which was established in a pilot study. During Sal exercise, pH decreased from 7.42 ± 0.01 at rest to 7.07 ± 0.02 but only to 7.34 ± 0.02 (P < 0.05) during the Bic trial. Arterial Po 2 was reduced from 103.1 ± 0.7 to 88.2 ± 1.3 Torr during exercise with Sal, and this reduction was not significantly affected by Bic. Arterial O2 saturation was 97.5 ± 0.2% at rest and decreased to 89.0 ± 0.7% during Sal exercise but only to 94.1 ± 1% with Bic (P < 0.05). Arterial Pco 2 was not significantly changed from resting values in the last minute of Sal exercise, but in the Bic trial it increased from 40.5 ± 0.5 to 45.9 ± 2.0 Torr (P < 0.05). Pulmonary ventilation was lowered during exercise with Bic (155 ± 14 vs. 142 ± 13 l/min;P < 0.05), but the exercise-induced increase in the difference between the end-tidal O2 pressure and arterial Po 2 was similar in the two trials. Also, pulmonary O2 uptake and changes in muscle oxygenation as determined by near-infrared spectrophotometry during exercise were similar. The enlarged blood-buffering capacity after infusion of Bic attenuated acidosis and in turn arterial desaturation during maximal exercise.

  • arterial O2 pressure
  • lactate
  • pH
  • rowing
  • ventilation

exercise-induced arterial hypoxemia is manifested as a reduced arterial O2 partial pressure (PaO2; Refs.1, 14, 19, 27,48, 49) and also, in some athletes, as a reduced saturation of Hb (SaO2; Refs.17, 23, 37, 51). Insufficient increase in ventilation (14), a pulmonary diffusion limitation (14), and ventilation-perfusion inequality (21, 26) may contribute to reduce PaO2. A low PaO2reduces SaO2, especially in acidic conditions because this affects the affinity of O2 to Hb as demonstrated by Bohr et al. (5). On the other hand, even a small increase in the inspired O2 fraction restores SaO2 during maximal exercise (40, 41,46). Temperature also affects the O2 dissociation curve, and both temperature and acidosis are likely to be of importance during maximal exercise. Thus aggravated acidosis, led by accumulation of lactic acid in blood combined with a limited availability of bicarbonate, reduces SaO2 during exercise (39).

An increase in blood pH is demonstrated during exercise after ingestion of bicarbonate with a subsequent increase in performance (34, 36,56, 57), but it is also reported that bicarbonate does not affect performance (2, 8, 44, 45) even during exercise at simulated altitude (29, 35). On the other hand, in the exercising rat, bicarbonate treatment moderates acidosis and largely attenuates the decrease in SaO2(20).

An effect of bicarbonate on SaO2 during sea level exercise in humans has not been determined. We evaluated the influence of acidosis on SaO2 during maximal rowing, which is associated with a marked reduction in both PaO2 and SaO2(22, 40, 41, 47). First, the blood temperature response to maximal rowing and the dose of bicarbonate that would attenuate acidosis were established in pilot studies. The main study evaluated the effect of a high dose of bicarbonate on arterial blood-gas variables, pulmonary gas exchange, and changes in muscle oxygenation as determined by near-infrared spectrophotometry (NIRS).


Five competitive oarsmen (Table 1) participated in the study after informed consent as approved by the Ethics Committee of Copenhagen (KF 01-280/98). No subject had any disease or injury in the 3 wk before the experiment, and they were not taking any medication. The subjects were not allowed to eat or to drink after midnight on the day of the experiment, which began at 0800.

View this table:
Table 1.

Individual variables for subjects participating in pilot studies and in the definitive study with indication of the dose of NaHCO3 and saline used in the trials

Exercise was performed on a rowing ergometer (model C; Concept II, Morrisville, VT). First, the subjects rowed for 12 min at work rates increasing from 150 to 250 W in steps of 50 W every third minute (warm-up). Thereafter they rowed for 5 min at an individually determined pace including several strokes at maximal intensity. After 5 min of recovery, a 2,000-m all-out time trial simulated an on-water competition. All rowers were familiar with this type of exercise from their off-season training, and they had a personal record for the distance known to have a variability of <1% (41). An all-out ergometer rowing protocol elicits a maximal O2uptake (V˙o 2 max) similar to that attained during cycling (13).

In the first pilot study (n = 6), temperature in the superior vena cava was recorded with a 132F5 catheter inserted through the basilic vein of the nondominant arm. The temperature is presented as the mean over the last minute of each submaximal work rate, every 15 s during the maximal row and for the first 5 min of the recovery. The temperature increased from the onset of exercise to reach 38.9 ± 0.1°C at the end of maximal rowing (Fig.1).

Fig. 1.

Blood temperature (vena caval) at rest, during a warm-up (W-u) at 150, 200, and 250 W, and before and during maximal ergometer rowing and 30 min into the recovery (n = 6). Values are means ± SE with the upper and lower range. ●, Different from rest, P < 0.05.

The second pilot study (n = 8) established the dose of bicarbonate that would attenuate acidosis during maximal rowing. Subjects received in a random double-blind fashion either sodium bicarbonate (1 M; 100–325 ml; Table 1) or an equal volume of isotonic saline in a crossover study design with 7 days between the two trials. However, infusion of sodium bicarbonate increased the concentration of sodium in blood to a higher level than with isotonic saline, which might affect plasma volume and in turn O2transport.

In the main study (n = 5), we aimed at a dose of sodium bicarbonate that was expected to eliminate acidosis during maximal exercise. Furthermore, with administration of an equal volume of 1 M saline in the placebo condition, we expected that the level of blood sodium, and in turn plasma volume, would be the same in the two trials. Treatments were applied by randomization in a double-blind fashion in a crossover study design with 7 days between trials. One subject received only 200 ml of sodium bicarbonate because of catheter failure and, accordingly, also 200 ml of saline. In both trials, the subjects experienced headache in the first minute of the recovery.

A catheter (1.0 mm, 20 gauge) was inserted in the radial artery of the nondominant arm. Infusions of bicarbonate or saline were administered through a central catheter (1.7 mm, 16 gauge) inserted in an upper arm vein. The total dose of sodium bicarbonate or saline to be infused was divided into 60-ml syringes emptied at a constant rate (∼50–60 ml/min) according to the expected race time.

Arterial blood samples were obtained anaerobically in heparinized syringes (4042E, SIMS) at rest and at the end of each submaximal intensity during the warm-up, every minute during the maximal row, and at minutes 1, 2, 3, 4, 5, 10, 20, and 30 of the recovery. Samples were kept on ice and analyzed for blood-gas variables, Hb, glucose, sodium, calcium, potassium, and lactate in plasma by use of an ABL 615 apparatus (Radiometer, Copenhagen, Denmark) with co-oximetry for determination of SaO2. Blood gases were corrected to the average blood temperature established for each time point (Fig. 1). For the subjects who participated in the temperature pilot study, the individual temperature change was used. Thus we did not take into account the extent to which infusion of sodium bicarbonate or saline would affect blood temperature. However, even with the unlikely assumption that the effect is limited to blood only, the decrease in blood temperature is estimated to be <0.1°C. The O2 content in arterial blood was calculated as the sum of bound and dissolved O2.

Ventilatory variables were determined by use of an integrated system (MedGraphics 2001; Medical Graphics, St. Paul, MN) with values stored on a hard disk. The O2 was determined electrochemically, and CO2 was evaluated by an infrared analyzer. O2 uptake (V˙o 2), ventilation (V˙e), respiratory frequency, tidal volume (Vt), expired CO2(V˙co 2), and end-tidal partial pressures for O2 (Pet O2) and CO2 (Pet CO2) were averaged for every 15 s and presented as the mean for the last minute of exercise. The mixed expired CO2 fraction was also measured for calculation of the mixed expired CO2 pressure whereby the dead space ratio (Vd/Vt) was estimated in the last minute of exercise. The increase in plasma sodium may have an influence on respiration; however, a previous study (50) has shown that plasma volume expansion does not affect pulmonary gas exchange. After each exercise trial, the subjects were asked to express their perceived exertion as guided by a visual scale (6).

The concentration changes of deoxygenated (ΔHb) and oxygenated Hb (ΔHbO2) of the right vastus lateralis muscle were assessed by NIRS (3, 7, 9, 32). A continuous-wave photometer was used (NIRO500; Hamamatsu Phototonics, Hamamatsu, Japan) with light transmitted via a fiber-optic cable and reflected light delivered via a second cable to a photomultiplier operating at four wavelengths. From the measured optical densities, the chromophor concentration change in microliter per liter tissue was calculated by using computer software (ONMAIN; Hamamatsu). The algorithm is based on a modified Lambert Beer's law: A = α · c · d · B+G, where A is the measured attenuation in optical density, α is the specific extinction coefficient of the absorbing compound (μM/cm), c is the concentration of the absorbing compound (μM),d is the distance between the optodes on the skin surface (4 cm), B is the differential pathlength factor, andG is a factor introduced to account for scattering of light in the tissue. The differential pathlength factor adopted for the leg was 4.94 (16). Data were obtained over 5 s, and variables are presented as the average during the last minute of maximal rowing. Individual and intramuscular variations in Binfluence the estimate of · Hb and · HbO2. However, changes were evaluated over time with each subject being his own control. Movement associated disturbances of the NIRS signal over the leg were considered to be the same in the two trials.

Data are presented as means ± SE. Results were evaluated by ANOVA for repeated measures and by two-tail Student's t-test for paired data. Statistical significance was set at the 95% confidence limit (P < 0.05).


The race time was faster and the expressed perceived exertion was lower with sodium bicarbonate than in response to saline exercise [median of 6 min 21 s (range 6 min 16 s to 6 min 58 s) vs. 6 min 28 s (6 min 23 s to 7 min 10 s) and median of 18 (range 13–19) vs. 19 (17–19), respectively;P < 0.05].

Lactate, pH, and bicarbonate.

During the warm-up, the concentration of lactate in arterial blood remained below 5 mM and then increased progressively during maximal exercise with saline (Fig. 2). The level of lactate remained high in the first minutes of the recovery but then decreased toward the preexercise level in both the sodium bicarbonate and saline trials. However, in the sodium bicarbonate trial the concentration of lactate was higher from the fourth minute of exercise and throughout the 30 min of the recovery compared with the saline trial. Blood lactate increased by 10 ± 2 mM in the sodium bicarbonate compared with saline trial, and in one subject during the sodium bicarbonate trial the concentration of lactate reached 32 mM.

Fig. 2.

Individual pH, arterial O2 pressure (PaO2), lactate concentration, Hb O2 saturation (SaO2), arterial CO2 pressure (PaCO2), and the concentration of bicarbonate in arterial blood (HCO3 ) at rest, during W-u, and in response to a 2,000-m ergometer row (maximal) (n = 5). Values are for infusion of saline (○) and for the administration of sodium bicarbonate (●). Infusions were made constantly throughout maximal exercise.

The concentration of blood bicarbonate was not significantly affected during the warm-up, but it became markedly reduced during maximal exercise. A further reduction was seen in the first minutes of the recovery. Thereafter, blood bicarbonate increased but remained below the preexercise level. With the infusion of sodium bicarbonate, blood bicarbonate was reduced only at the third minute of exercise and therefore remained higher than in the control trial (Fig. 2).

In response to exercise, pH decreased to reach the lowest level in the last minute of maximal exercise. After exercise, pH recovered but remained below the resting level. With infusion of sodium bicarbonate, pH was not significantly reduced and it remained higher than in the control trial.

Blood-gas variables.

The PaO2 decreased only at the last incremental stage of the warm-up, whereas SaO2 was not significantly changed from the resting level (Fig. 2). PaO2 was reduced from the onset of maximal exercise, and SaO2 decreased progressively to reach a lowest value in the last minute. Resaturation was established within the first minute of the recovery. During exercise, the sodium bicarbonate trial did not significantly affect the reduction in PaO2 compared with saline. In contrast, SaO2 was improved during sodium bicarbonate-supplemented exercise.

During the saline trial, PaCO2 decreased only in first minute of exercise (Fig. 2). In the last minutes of rowing with sodium bicarbonate, PaCO2 was higher than during exercise with saline. The concentration of Hb was not changed in response to maximal exercise (Table 2). Thus CaO2 decreased to the same extent in both trials.

View this table:
Table 2.

Blood variables and near-infrared spectrophotometry determined changes in muscle oxygenation during maximal ergometer rowing with infusion of saline or NaHCO3

The concentration of glucose and potassium increased during rowing with no significant effect of sodium bicarbonate (Table 2). In contrast, the plasma calcium concentration increased only during exercise with saline. During exercise, the increase in sodium increased to the same extent in both trials.


Muscle oxygenation remained stable at rest, but from the onset of maximal exercise, ΔHb and ΔHbO2 increased and decreased, respectively (Fig. 3, Table2). These exercise-induced changes in muscle oxygenation were not significantly affected by sodium bicarbonate.

Fig. 3.

Near-infrared spectrophotometry-determined concentration change of oxygenated (ΔHbO2) and deoxygenated Hb (ΔHb) in the vastus lateralis muscle during maximal ergometer rowing and into the recovery. Representative values from 1 subject.

Ventilation and heart rate.

Heart rate increased similarly during maximal exercise with saline and with sodium bicarbonate (Table3). During control exercise, respiratory frequency, Vt, V˙e,V˙co 2, V˙o 2, Pet O2, and the respiratory exchange ratio increased. The Pet O2-PaO2difference was widened, whereas Pet CO2 decreased. Furthermore, Vd/Vt decreased in response to maximal exercise with saline (from 0.39 ± 0.03 at rest to 0.20 ± 0.02;P < 0.05).

View this table:
Table 3.

Heart rate and ventilatory variables in response to a 2,000-m maximal ergometer row with infusion of saline or NaHCO3

During exercise with sodium bicarbonate, Pet CO2 increased to above the level established during exercise with saline (Table 3).V˙co 2 was also higher with sodium bicarbonate than during exercise with saline. BecauseV˙o 2 was not significantly affected by sodium bicarbonate, the respiratory exchange ratio increased. Exercise-induced change in Pet O2and the Pet O2-PaO2difference were not significantly affected by sodium bicarbonate. In contrast, V˙e reached a lower level during exercise with bicarbonate, whereas the increased Vt and respiratory frequency during exercise were not affected by bicarbonate. The Vd/Vt was also reduced during exercise with sodium bicarbonate (0.33 ± 0.04 at rest vs. 0.12 ± 0.03 during exercise; P < 0.05), and this reduction tended to be even lower than during exercise with saline (P = 0.09).


During maximal exercise, bicarbonate infusion attenuated acidosis whereby SaO2 increased, supporting the theory that the Bohr effect contributes to exercise-induced arterial desaturation when PaO2 is low. The increase in SaO2 during bicarbonate-supplemented exercise did not affect pulmonary V˙o 2 or changes in muscle oxygenation. Infusion of sodium bicarbonate did result in a small improvement in performance, a lowered pulmonary ventilation, and a marked increase in blood lactate.

Blood-gas variables.

The exercise-induced reduction in SaO2 can be attributed to factors that are known to influence the O2dissociation curve. In particular, PaO2 is critical for SaO2 in the case in which acidosis develops secondary to the pronounced accumulation of lactic acid in blood, and there is limited availability of blood bicarbonate. On the other hand, during exercise with sodium bicarbonate, the concentration of blood bicarbonate was maintained close to the resting level, i.e., there was enough blood bicarbonate to eliminate excess hydrogen ions. In this case, SaO2 was little affected by the decrease in PaO2.

Another factor that influences the Hb affinity to O2 is temperature, which should be considered in the calculation of SaO2. In the present study, SaO2 is measured by an ABL 615 analyzer, in which determination of SaO2 is independent of small changes in temperature. Thus the observed desaturation in our study is comparable with data from previous studies evaluating hypoxemia during maximal rowing, in which the results were reported at 37°C (22, 40,41). In addition to SaO2, the increase in blood temperature affects the partial pressure of blood gases. The present study found that PaO2 is higher than that found during other rowing studies (22, 40, 41). This could be because the other studies did not correct for hyperthermia in the analysis of their blood gases.

The use of an average temperature rather than individual temperature changes may have introduced a certain degree of inaccuracy in the assessment of PaO2. However, rowing-induced hyperthermia appears to be a consistent finding among subjects (Fig.1). Erythrocyte 2,3-diphospho-d-glycerate can also affect SaO2. Although such an effect was not evaluated in the present study, the 2,3-diphospho-d-glycerate concentration appears to be unchanged during maximal exercise (52, 54).

The mechanisms proposed to reduce PaO2 during exercise (15) were not evaluated separately in the present paper. We did observe marked variations in PaCO2 among the subjects during the all-out row (Fig. 4). In one subject, PaCO2 was close to 30 Torr, whereas hypoventilation was manifest in others with PaCO2 at, or above, 40 Torr during exercise with infusion of saline. Furthermore, as in previous studies (40,41), the Pet O2-PaO2difference increased during maximal rowing, indicating ventilation-perfusion inequality and diffusion limitation.

Fig. 4.

O2 pressure (PaO2) related to the arterial CO2 pressure (PaCO2). Values obtained at each minute during maximal rowing (saline trial): ○, subject 2; ▵, subject 4; ▿, subject 8; □, subject 10, and ◊,subject 11.

Maximal rowing elicits a cardiac output in excess of 30 l/min (41), and the subject with the lowest PaO2 also had the lowest PaCO2, suggesting a pulmonary limitation to O2 transport rather than insufficient breathing in this individual. Thus hypoventilation, ventilation-perfusion mismatch, and a fast blood transit time may all have contributed to reduce PaO2 during maximal rowing.


Breathing appears entrained to the rowing rhythm (31). Both peripheral and central factors influence the control of ventilation during exercise. In the present study, potassium increased to the same extent in the two trials, whereas sodium bicarbonate increased SaO2 and pH. Oren et al. (43) found that bicarbonate treatment also tended to reduce the increase in V˙e during cycling. In that study, hyperoxia slowed the ventilatory kinetics to a greater degree during acidosis than during control or alkalosis, indicating the influence of carotid bodies on respiration. Previously, our laboratory demonstrated that hyperoxia does not affect the ventilatory response to maximal rowing (40, 41). Thus, in the case of maximal rowing, a likely explanation for the reduction in V˙ewith bicarbonate infusion is that a central influence of excess H+ concentration on respiration is attenuated.

With the lowered ventilatory response during exercise with sodium bicarbonate, we expected a decrease in PaO2. From Table 3 it appears that the average Pet O2 tended to be lower in the bicarbonate trial. This can explain why the widened Pet O2-PaO2difference also tended to be lower during exercise with sodium bicarbonate than with saline. Nevertheless, these trends did not reach statistical significance. In the view of the reducedV˙e, the reason that sodium bicarbonate did not change PaO2 may relate to reduced Vdventilation.


Extracellular acidosis may depress muscle contraction (30) and muscle glycogen utilization (53) and provoke fatigue (10). In fact, preexercise administration of bicarbonate increases the torque in an isometric contraction (56). Furthermore, an increase in extracellular bicarbonate leads to a higher efflux of protons (increased lactate output) from skeletal muscle (25, 38).

The release of lactate from cells is a pH-sensitive lactate-proton translocation in which a low external pH supports its release from the muscle (33) and its uptake by, e.g., erythrocytes, kidney, liver, muscle, and the brain (28). However, muscle lactate release involves not only the lactate-proton transport but also diffusion via bicarbonate-chloride exchange and via sodium-hydrogen exchange (28). Transmembrane transport of bicarbonate also occurs via a sodium-dependent bicarbonate cotransport (58). This could be important for an increase in intracellular bicarbonate when sodium bicarbonate induces an excess extracellular buffer capacity. The result would be an increase in net release of lactate supported by the finding that intracellular pH may increase during exercise after ingestion of bicarbonate (12).


NIRS is used to evaluate muscle O2 extraction during muscular activity (3, 4, 7, 31, 42), and maximal rowing affects light absorption related to an increase in muscle Hb (9) and a decrease in oxygenated Hb (41).

In the present study, NIRS evaluated whether an increase in SaO2 affected the level of muscle oxygenation and in turn muscle O2 delivery. However, although small changes in SaO2 result in increased cerebral oxygenation (18, 41), muscle oxygenation did not appear to be affected by the increased SaO2 during exercise with bicarbonate infusion. In support, with a 10% increase in CaO2 when hyperoxia restores arterial desaturation during rowing (40), the NIRS-determined muscle oxygenation was not different from that during exercise in normoxia (41). These data indicate that there is no significant effect of exercise-induced hypoxemia on muscle O2 delivery.

The use of NIRS is based on the assumption that the small area of muscle evaluated reflects changes in the whole muscle and that motion-induced changes in the scattering of light were similar in the two trials. Furthermore, changes in skin blood flow are not considered to affect the muscle NIRS recordings (32). Another important bias is that changes in Hb O2 status may not be distinguished from that of myoglobin (55).

The present data do not suggest that the increase in performance with the infusion of sodium bicarbonate is explained by an increase in O2 extraction or in V˙o 2. The effect of sodium bicarbonate on ionic calcium reflects the binding to plasma proteins (59) and may not affect muscle contraction. The reduced V˙e with sodium bicarbonate could affect the level of respiratory muscle work of consequence for work capacity (24). It is also considered that the small changes in performance and perceived exertion may reflect that fatigue is related to intracellular pH (28).

We conclude that infusion of sodium bicarbonate attenuates the rise in pulmonary ventilation during maximal exercise, supporting a role of pH for ventilatory control. An enhanced Hb affinity to O2binding did not appear to impede muscle O2 extraction. Rather, performance increased with an even higher concentration of blood lactate. During maximal exercise with a marked reduction in the arterial O2 pressure, a reduction in pH affects the arterial O2 saturation of Hb.


P. P. Bredmose received a grant from The Købmand i Odense Johann and Hanne Weimann f. Seedorfs foundation. S. Volianitis received a Marie Curie fellowship (HPMF-CT-2000-00526) from the European Union.


  • Address for reprint requests and other correspondence: H. B. Nielsen, Dept. of Anesthesia 2041, Rigshospitalet 2041, Blegdamsvej 9, DK-2100 Copenhagen Ø, Denmark (E-mail:h.bay{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

  • 10.1152/japplphysiol.00398.2000


View Abstract